Managerial Algorithmics™ is a basic discipline for understanding managerial competence and performance on the basis of an accurate computational model of perceiving, emoting, thinking and doing. It rests on the notion that humans can learn from computers (or, more accurately, from the humans that have designed and interacted with computers) as much as computer designers can learn from humans whose ‘minds’ they are trying to get computers to emulate. Intelligent artificiality (as opposed to artificial intelligence) posits that the language of algorithms and computational complexity analysis can inform the way humans interact with the problems that they face by forcing them to be more precise in the ways in which they formulate problems and more adaptive in the ways in which they generate and test solutions. The methodology of Managerial Algorithmics as applied to business problem solving rests on three steps:
Step 1. Represent everyday problems as formal problems, i.e. a set of initial conditions, a set of desired conditions and a search space of possible paths that connect the initial to the final conditions;
Step 2. Measure the complexity of the process by which the space of possible solutions is searched, the probability that the search will converge to a solution of a minimum quality and estimate the time required to execute this search;
Step 3. Choose the solution generating process (the search process) that has the optimal value metric (highest probability of convergence within a certain time at a certain level of accuracy, for instance).
The problem complexity measures used in Managerial Algorithmics are adapted from the theory of computation (time complexity) and from information theory (Kolmogorov complexity). The ‘art’ of the managerial algorithmics is that of representing everyday problems, usually adumbrated in a loose, fuzzy and uninformative language, into the unforgiving language of algorithms. The promise of managerial algorithmics is that of turning individuals, groups and organizations taken as a whole into better problem framers and solvers. ‘Better’ and not ‘best’; and ‘framers and solvers’, not merely ‘solvers’ – for it is ‘puzzles’ that one ‘solves’; ‘problems, on the other hand, must be produced, refined and sometimes dissolved before one can speak of ‘solving’ them.
Organizational Algorithmics (TM) extends the basic idea behind intelligent artificiality from the individual to the organizational levels, by pulling off a common ontological trick and modeling organizations themselves as large scale computational devices. What organizations do is modeled by the running of these devices. The plans that what organizations do implement is modeled as the algorithms that run on the computational devices. Computational problems model the problems that organizations solve when they do what they do. And ‘data’ in the real world is ‘data’ in the ersatz world of the computational model. Organizational Algorithmics takes seriously the metaphor of ‘the universe as a computational device, elaborated by Seth Lloyd (link to website) and Stephen Wolfram (link to website) and popularized by Rudy Von Braun Rucker (link to website), and makes it possible to design and implement organizational processes, procedures, routines and structures that make organizations into more effective, efficient or adaptive problem solvers.
Together, Managerial Algorithmics and Organizational Algorithmics were used by Mihnea Moldoveanu to design the organizational production functions of both of the companies he has built, Hefaistos, Inc. and Redline Communications Group, Inc (link to ‘Companies built’ in this website). Unlike the majority of social science models that are to generate explanations and justifications, they have been used predictively and performatively, as design and forecasting instruments.